Commit message (Collapse) | Author | Age | |
---|---|---|---|
* | Go: Update generated wrapper functions for TensorFlow ops. | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160341769 | ||
* | Add checkpoint conversion for models that use the attention mechanism ↵ | A. Unique TensorFlower | 2017-06-27 |
| | | | | | | implemented in tensorflow/contrib/legacy_seq2seq/python/ops/seq2seq.py. PiperOrigin-RevId: 160340994 | ||
* | Update ops-related pbtxt files. | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160340888 | ||
* | Fix bugs in ScatterNd and add ScatterNdNonAliasingAdd. | Eugene Brevdo | 2017-06-27 |
| | | | | | | | | | tf.scatter_nd_non_aliasing_add acts similarly to tf.scatter_nd_add but works on non-ref objects (i.e., Tensors -- not Variables). This means it has a gradient with respect to the primary input as well as the updates. It does its best to avoid making extra copies of the input. PiperOrigin-RevId: 160339328 | ||
* | Internal change. | Dandelion Man? | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160339087 | ||
* | Add note about `tf.test.mock` to docs for `tf.test` | Mark Daoust | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160338811 | ||
* | Add name_scopes to mnist_deep.py for a cleaner graph layout. | Mark Daoust | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160338775 | ||
* | Minor cleanup: remove unused dependencies and inclusions | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160334030 | ||
* | Make ResourceHandle not be a proto | Geoffrey Irving | 2017-06-27 |
| | | | | | | | I'm trying to make core/kernels independent of protos. Currently the dtype ResourceHandle is itself a proto. After this CL, ResourceHandle is a normal C++ type which gets converted to/from ResourceHandleProto at (de)serialization time. RELNOTES: n/a PiperOrigin-RevId: 160329002 | ||
* | Readd the new tensors and variables documents, with tests passing. | Alexandre Passos | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160324191 | ||
* | Moves TensorCApi from c_api.cc to c_api_internal.h, where it can be used | A. Unique TensorFlower | 2017-06-27 |
| | | | | | | by other code that require access to the underlying TensorBuffers. PiperOrigin-RevId: 160323362 | ||
* | tfdbg: revert dns:/// prefix in gRPC mode | Shanqing Cai | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160319348 | ||
* | Add a time estimation to HloCostAnalysis and represent properties as a map ↵ | Bjarke Hammersholt Roune | 2017-06-27 |
| | | | | | | so that adding more properties will be easier, e.g. in a sub-class. PiperOrigin-RevId: 160318494 | ||
* | tfdbg doc: fixes and improvements | Shanqing Cai | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160318411 | ||
* | TESTFIX: distributions:trig_test wasn't passing in ASAN mode. | Ian Langmore | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160315597 | ||
* | Automated g4 rollback of changelist 159454657 | Yangzihao Wang | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160314706 | ||
* | Migrate TensorForest v4 python to contrib. | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160308805 | ||
* | Update path to print_selective_registration_header.py in comment | Brian Williammee | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160308173 | ||
* | Add more tests for BatchNormTraining. | A. Unique TensorFlower | 2017-06-27 |
| | | | | | | RELNOTES: n/a PiperOrigin-RevId: 160307959 | ||
* | Use C API to implement Operation._output_types | A. Unique TensorFlower | 2017-06-27 |
| | | | | | | | This change first converts the _output_types member to a property and then implements it using C API if it is enabled. PiperOrigin-RevId: 160306227 | ||
* | Updates more text classification examples in examples/learn. | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160305131 | ||
* | Added more reduce window tests. | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160301509 | ||
* | [XLA] Remove "operand" argument from HandleReducePrecision. | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160301461 | ||
* | Cudnn compatible LSTMCell and LSTMBlockCell | James Qin | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160300668 | ||
* | Add a functionality of remote fused graph transformation to fuse graphs by ↵ | A. Unique TensorFlower | 2017-06-27 |
| | | | | | | op type PiperOrigin-RevId: 160300039 | ||
* | [TF:XLA] Reject operators with resource outputs on CPU and GPU devices. | Peter Hawkins | 2017-06-27 |
| | | | | | | We were checking for resource inputs but not resource outputs, which led to accidental fusion of some TensorArray ops on CPU and GPU. PiperOrigin-RevId: 160294302 | ||
* | Shrink the test size to avoid OOM error on old GPUs. | Yangzihao Wang | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160292834 | ||
* | Fix documentation of Estimator class (invalid quotes). | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160292803 | ||
* | [SE] ThenConvolveWithAlgorithm vlogs algorithm configs. | Jingyue Wu | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160292762 | ||
* | tfdbg example: fix --tensor_size issue in debug_fibonacci | Shanqing Cai | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160290541 | ||
* | [TF:XLA] Update Tensorflow LLVM release to upstream r306370. | Peter Hawkins | 2017-06-27 |
| | | | | | | Fix broken XLA build. PiperOrigin-RevId: 160284588 | ||
* | Fix uninitialized values in TensorForest code. | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160284420 | ||
* | [TF:XLA] Refactor the sigmoid op as a rescaled tanh. | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160282472 | ||
* | [XLA] Remove the "hlo dumper" parameter of xla::Compiler and its piping. | Eli Bendersky | 2017-06-27 |
| | | | | | | | | | This dumper is no longer necessary since the restructuring of HLO dumping and the addition of MaybeDumpHloModule which heeds to the right flags. The remaining bits didn't have additional functionality, but constituted a lot of boilerplate that has to be propagated throughout the backends. PiperOrigin-RevId: 160281798 | ||
* | Clean out the config and params in TPUEstimator. | Jianwei Xie | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160281507 | ||
* | Fixed the bug that Estimator does not make deepcopy of params in constructor | Jianwei Xie | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160281247 | ||
* | Automated g4 rollback of changelist 160278032 | Geoffrey Irving | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160280961 | ||
* | Updates some more examples in examples/learn. | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160278757 | ||
* | [XLA] Add tuple support to Literal::CreateFromShape. | Peter Hawkins | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160278561 | ||
* | Don't include node_def.proto.h in node_def_util.h | Geoffrey Irving | 2017-06-27 |
| | | | | | | | The goal is to make kernels mostly independent of proto headers, which will let us lock down our .so imports. RELNOTES: n/a PiperOrigin-RevId: 160278032 | ||
* | Add RWSE (root weighted squared error) to the WALS estimator. | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160276937 | ||
* | Explicitly use "dns" URI scheme when using DNS names or literal IP | A. Unique TensorFlower | 2017-06-27 |
| | | | | | | | addresses with gRPC. This avoids problems in environments in which the default URI scheme is something other than "dns". PiperOrigin-RevId: 160276862 | ||
* | Make sure all convolution tests are testing non-trivial cases, i.e. where ↵ | A. Unique TensorFlower | 2017-06-27 |
| | | | | | | | | | not all inputs are 0, leading to an all-0 output, which masks most possible bugs. We do not check-fail on 0-sized dimensions as tests for these special cases exist. PiperOrigin-RevId: 160274593 | ||
* | [TF:XLA] Inline the sigmoid operation instead of mapping it elementwise. | A. Unique TensorFlower | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160274436 | ||
* | Automated g4 rollback of changelist 155070869 | Peter Hawkins | 2017-06-27 |
| | | | | PiperOrigin-RevId: 160249526 | ||
* | Lets the HParam parser also accept True and False as inputs, since that's ↵ | A. Unique TensorFlower | 2017-06-26 |
| | | | | | | how python prints booleans. PiperOrigin-RevId: 160234658 | ||
* | Automated g4 rollback of changelist 160196550 | A. Unique TensorFlower | 2017-06-26 |
| | | | | PiperOrigin-RevId: 160222942 | ||
* | Automated g4 rollback of changelist 160196566 | A. Unique TensorFlower | 2017-06-26 |
| | | | | PiperOrigin-RevId: 160222930 | ||
* | [XLA] Simplify the fusion heuristic | David Majnemer | 2017-06-26 |
| | | | | | | | | | | | | We had two different aspects of the fusion heuristic: - Don't fuse a producer into a consumer if there exists a path from the producer to the consumer which cannot be fused. - Don't fuse a producer into a consumer if any consumer of the producer cannot fuse. These can be combined into one, simpler, heuristic. PiperOrigin-RevId: 160222771 | ||
* | Support advisor in all places (Command line, APIs) | A. Unique TensorFlower | 2017-06-26 |
| | | | | | | Add expensive operation checker PiperOrigin-RevId: 160222348 |